From 0b8659bf3ff698a4cc6af9bb946a1e25bacc9aae Mon Sep 17 00:00:00 2001
From: Joel <me@joelgustafsson.com>
Date: Thu, 30 Aug 2018 17:43:21 +0200
Subject: [PATCH] Remove latent rotate point cloud from transforms

Has been in augmentors for a while
---
 action_recognition/transforms/__init__.py     |  1 -
 .../transforms/rotate_point_cloud.py          | 88 -------------------
 2 files changed, 89 deletions(-)
 delete mode 100644 action_recognition/transforms/rotate_point_cloud.py

diff --git a/action_recognition/transforms/__init__.py b/action_recognition/transforms/__init__.py
index 5cf88ba..24ce3e6 100644
--- a/action_recognition/transforms/__init__.py
+++ b/action_recognition/transforms/__init__.py
@@ -15,4 +15,3 @@
 from .speed import Speed
 from .extract_keypoints import ExtractKeypoints
 from .interpolate_keypoints import InterpolateKeypoints
-from .rotate_point_cloud import RotatePointCloud
diff --git a/action_recognition/transforms/rotate_point_cloud.py b/action_recognition/transforms/rotate_point_cloud.py
deleted file mode 100644
index 0fa27b2..0000000
--- a/action_recognition/transforms/rotate_point_cloud.py
+++ /dev/null
@@ -1,88 +0,0 @@
-import numpy as np
-from sklearn.base import BaseEstimator, TransformerMixin
-
-
-class RotatePointCloud(BaseEstimator, TransformerMixin):
-    """Rotates point clouds for data augmentation.
-
-    Use with care, as you will also need to add corresponding labels to the
-    added points.
-
-    Parameters
-    ----------
-    number_of_added_points : int, optional
-        The number of augmented points to add.
-
-    """
-
-    def __init__(self, number_of_added_points=2):
-        self.number_of_added_points = number_of_added_points
-
-    def fit(self, X=None, y=None, **fit_params):
-        """Returns self unchanged, as there are no parameters to fit.
-
-        Parameters
-        ----------
-        X : ignored
-        y : ignored
-        fit_params : ignored
-
-        Returns
-        -------
-        self : unchanged
-
-        """
-        return self
-
-    def transform(self, chunks):
-        """Rotates the inputted point clouds randomly.
-
-        Does not transform to origin, so take care if your data isn't
-        already centered.
-
-        Parameters
-        ----------
-        chunks : array-like, shape = [n_chunks, n_keypoints, 3]
-
-        Returns
-        -------
-        data : array-like
-            shape = [n_chunks * (1 +number_of_added_points), n_keypoints, 3]
-
-        """
-
-        data = np.array([self._random_rotation(chunk)
-                         for chunk in chunks
-                         for i in range(self.number_of_added_points)])
-        return np.append(data, chunks, axis=0)
-
-    def _random_rotation(self, chunk):
-        rotated_chunk = np.copy(chunk)
-        rotation_matrix = self._rotation_matrix()
-
-        rotated_chunk = rotated_chunk @ rotation_matrix
-
-        return rotated_chunk
-
-    def _rotation_matrix(self):
-        tx = np.radians(np.random.random_integers(low=-180, high=180))
-        ty = np.radians(np.random.random_integers(low=-180, high=180))
-        tz = np.radians(np.random.random_integers(low=-180, high=180))
-
-        Rx = np.array([
-            [1, 0, 0],
-            [0, np.cos(tx), -np.sin(tx)],
-            [0, np.sin(tx), np.cos(tx)]
-        ])
-        Ry = np.array([
-            [np.cos(ty), 0, -np.sin(ty)],
-            [0, 1, 0],
-            [np.sin(ty), 0, np.cos(ty)]
-        ])
-        Rz = np.array([
-            [np.cos(tz), -np.sin(tz), 0],
-            [np.sin(tz), np.cos(tz), 0],
-            [0, 0, 1]
-        ])
-
-        return Rz